Investigating the relationship between glymphatic dysfunction and motor symptoms in Parkinson’s disease

Abstract

The glymphatic system is instrumental in cerebral waste drainage, and its impairment is mechanistically linked to the pathogenesis of neurodegenerative diseases like Parkinson's disease (PD). Nevertheless, the relationship between dysfunction in the clearance processes and the intensity of motor symptoms in PD remains inadequately characterized. This study applied the non-invasive Diffusion Tensor Image Analysis along the Perivascular Space technique to quantify the clearance network activity in PD patients compared with healthy controls (HCs), probing its prospects as an imaging biomarker for motor impairment. From January 2024 to June 2025, a total of 43 PD patients, diagnosed according to the clinical diagnostic criteria, along with 31 well-matched HCs, were enrolled. All participants underwent MRI, and the waste-clearance system function was evaluated using the DTI-ALPS index. Motor manifestations in the PD group were systematically rated using the MDS-UPDRS. An independent sample t-test was carried out to compare the disparities in this imaging metric between groups. Correlation and multiple regression analyses were further employed to explore the connections between the imaging metric and parkinsonian motor features. This study demonstrated a significantly lower MRI-derived metric in the PD group (1.19 ± 0.14) relative to HCs (1.27 ± 0.13; p = 0.019). The statistical significance of this difference persisted after covariance adjustment for age, years of education, and drinking history via ANCOVA (B = 0.080, 95% CI [0.015, 0.145], p = 0.016). Pearson correlation analysis revealed significant inverse associations between this radiographic measure and both the total MDS-UPDRS score (r = − 0.418, p = 0.005; Bootstrap 95% BCa CI [− 0.628, − 0.171]) and its motor subsection (Part III) score (r = − 0.424, p = 0.005; Bootstrap 95% BCa CI [− 0.644, − 0.169]). Multiple regression indicated that the imaging metric served as an independent negative predictor of motor symptom severity (standardized β = − 0.420, p = 0.010; B = − 42.45, Bootstrap 95% BCa CI [− 79.17, − 9.91]), accounting independently for 14.6% of the variability in motor severity (ΔR² = 0.146, p = 0.008). These results suggest an association between the imaging metric and the function of the clearance network in PD, as well as a significant correlation with motor disability. These findings warrant further investigation into the metric's potential as a neuroimaging marker for motor impairment.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40001-025-03125-x.

Introduction

Parkinson's disease (PD) exhibits a dual symptomatic structure,featuring central motor impairments along with numerous non-motor manifestations,which is recognized as a prevalent neurodegenerative condition [1]. This disease is defined by fundamental motor abnormalities, including bradykinesia, tremor at rest, muscle rigidity, and impaired postural stability [2, 3]. Beyond motor manifestations, patients with Parkinson’s disease commonly present various neuropsychiatric features including cognitive decline and mood disturbances [4]. Parkinson’s disease is classified as a type of synucleinopathy, characterized by the anomalous clumping of α-synuclein, which subsequently leads to the formation of Lewy bodies and Lewy neurites as its principal neuropathological features [5–7].The misfolded protein aggregation and dysfunction of its clearance mechanisms are considered pivotal in driving the progression of neuropathological alterations in PD [8–11].

Using in vivo imaging in animal models, researchers have indicated that dynamic course of cerebrospinal fluid (CSF) into and out of cerebral tissue relies on a transport network formed by astrocytic endfeet densely expressing AQP4 water channels [12].Anatomically situated within the perivascular spaces, the clearance pathway performs a critical function in supporting internal equilibrium of the central nervous system(CNS) by eliminating solutes and metabolic waste from cerebral tissue and facilitating their outflow into the meningeal lymphatic vessels and cervical lymphatic system [13, 14].

Recent neuroimaging and molecular studies have gradually elucidated that the glymphatic system, via its specialized perivascular pathways, may participate in regulating the clearance of neurotoxic proteins, such as β-amyloid, from the brain. Evidence from functional imaging and pathological research further indicates that dysfunction of this clearance mechanism is linked to the advancement of Alzheimer’s disease pathology [15, 16]. Pathological α-synuclein deposits could disrupt the asymmetric distribution of AQP4 water channels on astrocytic endfeet, thereby impairing glymphatic clearance efficiency. Such a vicious cycle might amplify aberrant protein aggregation and ultimately exacerbate Parkinson’s disease-like neuropathology [15, 17–20]. Therefore, evaluating the overall functional status of the clearance network could provide insights into the pathophysiological mechanisms of the disorder.

Neuroimaging techniques are instrumental in both the clinical diagnosis and mechanistic investigation of neurodegenerative disorders. Before the advent of perivascular space-based diffusion tensor imaging techniques for evaluating glymphatic function, invasive approaches such as intravenous or intrathecal injection of gadolinium-based contrast agents with subsequent clearance kinetic monitoring were predominantly used [21]. Although these approaches can yield valuable functional insights, the use of gadolinium chelates poses clinical limitations and potential risks of adverse effects [22].These technical limitations have motivated the emergence of non-invasive neuroimaging approaches. Notably, diffusion tensor imaging analysis along the perivascular space extends a novel approach to quantify the global activity of the glymphatic system in vivo, holding potential for early-stage detection of Parkinson’s disease. The seminal work by Taoka et al. proposed the DTI-ALPS index as a potential measure to assess the clearance network activity based on the dynamics of water diffusion within perivascular spaces [23]. Building upon this foundation, the present study aims to utilize this imaging metric to investigate glymphatic system function in Parkinson's disease patients and to evaluate the relationship between its functional index and the severity of motor symptoms.

Materials and methods

Informed consent

This research rigorously conformed to the ethical guidelines set forth in the Declaration of Helsinki (2013 updated version) and received formal authorization from the Institutional Review Board (IRB) of our medical institution (Authorization Number: EC-024-106; Authorization Date: January 20, 2024). Before taking part in the study, every participant signed a written informed consent form subsequent to being thoroughly briefed on the research aims, protocols, possible risks and advantages associated with magnetic resonance imaging examinations and scale evaluations, along with data privacy protection strategies.Cognitive capacity was evaluated for all participants, with Parkinson's disease patients undergoing independent decision-making capacity assessments by neurologists. Notably, informed consent documents were stored separately from research data, and participants retained the right to withdraw unconditionally at any time. For Parkinson's disease patients with cognitive impairments, additional informed consent was obtained from their legal proxies.

Study participants

This prospective study enrolled participants according to the MDS clinical diagnostic criteria [24]. Between January 2024 and June 2025, we recruited 43 PD patients at our institution. All PD patients were maintained on stable dopaminergic medication (levodopa equivalent dose: 300–600 mg/day) for at least 3 months. Exclusion criteria included: neurological comorbidities, drug-induced parkinsonism, major systemic diseases, MRI contraindications, and cognitive impairment (MMSE score < 18). Thirty-one age-, sex-, and education-matched healthy controls were concurrently enrolled, with exclusion criteria comprising: history of neurological/psychiatric disorders, family history of PD in first-degree relatives, and chronic use of psychoactive medications. All participants successfully completed MRI scans and standardized clinical assessments (see Fig. 1 for the screening flowchart).

Figure. 1.

The screening flowchart

Clinical assessments

This study implemented a standardized clinical assessment protocol for eligible participants who provided written informed consent. The standardized assessment protocol was implemented through synchronized evaluations using a fixed-interval weekly paradigm, with comprehensive data collection at two defined time points: baseline on enrollment day (Day 0) and endpoint (EP) at 7 ± 1 days post-enrollment. Motor symptoms were quantitatively assessed using MDS-UPDRS-III. Non-motor evaluations encompassed multidimensional assessments of cognitive function (MMSE) and neuropsychiatric status (HAMD; HAMA). To ensure data integrity, all scale assessments were performed by MDS-certified board-certified neurologists, with repeated measurements implemented for key metrics to verify reliability. Healthy control subjects were strictly selected based on three criteria: age matching, gender ratio matching, and absence of neurological disease history.

MRI acquisition and methods

The 3.0 Tesla magnetic resonance imaging system (GE SIGNA premier) employed in this study was configured with a high-density 48-channel neuroimaging coil. Participants were positioned in a head-first supine orientation, equipped with noise-attenuating earplugs and immobilized using foam padding for cranial stabilization. We acquired diffusion tensor imaging of the brain with these scanning parameters: TR 4000 ms, TE 85.8 ms, number of slices 54, slice thickness 3 mm, gradient directions 64, FOV 22 mm × 22 mm, matrix 128 × 128, b-values of 0 and 1000 s/mm². 3D T1WI employed these acquisition specifications: TR 7.2 ms, TE 3.0 ms, flip angle 8°, FOV 25.6 mm × 25.6 mm, matrix 256 × 256. For ROI localization, anatomical positioning was rigorously performed according to the internationally standardized ICBM-152 reference brain space template. To ensure the reproducibility of the research methodology and the reliability of the analytical results, all neuroimaging data analyses were independently performed by two board-certified and experienced neuroradiologists.

Image processing

The radiologists were blinded to the age, sex, group allocation, cognitive test performance, and clinical data when analyzing the images. We employed the FSL software suite (available at https://fsl.fmrib.ox.ac.uk/fsl/oldwiki) and the MRtrix3 software (accessible via https://mrtrix.readthedocs.io/en/latest/) to conduct preprocessing of diffusion tensor imaging data. This preprocessing pipeline encompassed a series of steps, namely noise reduction, Gibbs ringing artifact removal, distortion rectification, eddy current and motion artifact correction, skull extraction, tensor estimation, and non-linear spatial normalization to a reference atlas. Through this processing sequence, we obtained fractional anisotropy maps, along with directional diffusion coefficient maps corresponding to the three orthogonal axes (Dx, Dy, Dz). The DTI-ALPS index was computed as the quotient of the average x-axis diffusivity in both the projection (Dxproj) and association (Dxassoc) fiber regions divided by the average of the y-axis diffusivity in projection fibers (Dyproj) and the z-axis diffusivity in association fibers (Dzassoc). Mathematically, it can be represented by the formula: DTI-ALPS Index = mean(Dxproj, Dxassoc)/mean(Dyproj, Dzassoc) [23]. The specific details are shown in Fig. 2.

Figure.2.

(a) demonstrates the placement of two ROIs in regions corresponding to projection fibers, association fibers, while (b) illustrates the spatial relationship between the orientation of perivascular spaces and fiber directions

Statistical analysis

In the context of demographic data analysis, for continuous and categorical variables, group-wise comparisons were carried out employing independent samples t-tests and χ² tests, respectively. Quantitative variables that followed a normal distribution were analyzed by means of independent samples t-tests, whereas variables that did not conform to a normal distribution were inspected using the Mann–Whitney U test. Moreover, ANCOVA was used to evaluate the differences in this imaging index among the different groups. The multicollinearity among covariates was assessed through the variance inflation factor (VIF). For continuous variables with a normal distribution, Pearson's linear correlation analysis was applied; for data that did not follow a normal distribution, Spearman’s rank correlation was utilized. Confidence intervals and effect sizes were estimated by employing Bootstrap resampling (with 1000 samples and the BCa method). To clarify the impact of this computed imaging index on the motor status of patients with PD, a multiple linear regression model was constructed, with MDS-UPDRS-III scores serving as the dependent variable. All analyses were performed using SPSS Statistics 26 software (https://www.ibm.com/analytics/spss-statistics-software). Quantitative data are presented as mean ± standard deviation (Mean ± SD), and qualitative data are shown as percentages (%). The threshold for statistical significance was set at P < 0.05.

Power analysis and sample size estimation

Based on previously reported effect sizes of DTI-ALPS (Cohen’s d ≥ 0.8), we performed a pre-experimental power calculation with the aid of G*Power 3.1 software. An independent samples t-test (two-tailed, α = 0.05, power = 80%) indicated a minimum required total sample size of N = 52 (26 per group). To ensure adequate sensitivity for detecting medium effects (d ≥ 0.5), we ultimately enrolled 74 participants. Post hoc power analysis confirmed that the current sample achieves 87.94% statistical power for detecting d = 0.8, meeting the study's sample size requirements.

Results

Clinical characteristics

As presented in Table 1, intergroup comparisons showed no statistically significant disparities in gender distribution (p = 0.270), age (p = 0.562), years of education (p = 0.187), smoking status (p = 0.946), or modified Fazekas scores (p = 0.885). However, significant between-group differences were observed in cognitive function (p < 0.001), depressive symptoms (p < 0.001), anxiety levels (p < 0.001), and alcohol consumption status (p = 0.028).

Table 1.

Baseline clinical characteristics

	HC	PD	P
Participants	31 (41.89%)	43 (58.10%)	–
Sex, n (%) (Male/Female)	14 (45.16%)/17(54.84%)	25 (58.14%)/18(41.86%)	0.270
Age, y	66.10 ± 6.33	67.12 ± 8.13	0.562
Disease duration, y	–	5.03 ± 3.43	–
Hoeh & Yahr, y	–	2.19 ± 1.08	–
Years of education	8.19 ± 3.67	7.21 ± 2.70	0.187
Total MDS-UPDRS	–	57.63 ± 24.90	–
MDS-UPDRS III	–	30.65 ± 15.45	–
MMSE	29.10 ± 1.66	25.98 ± 4.72	< 0.001*
HAMD	2.16 ± 2.98	7.30 ± 6.67	< 0.001*
HAMA	1.19 ± 1.60	4.34 ± 4.47	< 0.001*
BMI	24.41 ± 2.71	24.49 ± 2.84	0.907
Smoking (Yes/No) n (%)	7 (22.58%)/24(77.42%)	10 (23.26%)/33(76.74%)	0.946
Alcohol use (Yes/No) n (%)	13 (41.94%)/18(58.06%)	8 (18.60%)/35(81.40%)	0.028*
DTI-ALPS	1.27 ± 0.13	1.19 ± 0.14	0.019*
Modified Fazekas	1.06 ± 0.36	1.05 ± 0.49	0.885

Normally distributed data were reported in the form of mean ± standard deviation

MDS-UPDRS: Movement Disorder Society-Unified Parkinson’s Disease Rating Scale; HAMD: Hamilton Depression Scale; HAMA: Hamilton Anxiety Scale; Hoeh & Yahr: Hoehn-Yahr staging; DTI-ALPS: Diffusion Tensor Imaging Analysis Along Perivascular Spaces; MMSE: Mini-Mental State Examination; Modified Fazekas: Modified Fazekas Scale for White Matter Hyperintensities

^*Represents statistical significance of p < 0.05

Imaging characteristics

The intra-rater reliability was confirmed by test-retest analysis (ICC 0.926, 95% CI 0.885–0.953).Statistical analyses revealed that PD patients exhibited significantly lower DTI-ALPS values compared to healthy controls. As illustrated in Fig. 3, the PD group demonstrated markedly reduced DTI-ALPS measurements (1.19 ± 0.14) versus controls (1.27 ± 0.13), with this difference reaching statistical significance (p = 0.019).

Figure. 3.

Box-and-whisker plot

An ANCOVA (Table 2) demonstrated that after adjusting for age, education level, and alcohol consumption status, healthy controls maintained significantly higher DTI-ALPS indices than PD patients (B = 0.080 units, 95% CI [0.015, 0.145], p = 0.016), with this association showing moderate effect size (partial η² = 0.081). Neither education duration (p = 0.070), alcohol consumption (p = 0.076), nor age (p = 0.115) showed significant effects. All variance inflation factors (VIF) for covariates (age, education duration, and alcohol consumption) remained below 2, indicating robust model stability and effectively ruling out potential multicollinearity interference (see Supplementary Table 1 for details).

Table 2.

Effects of Group and Covariates on DTI-ALPS in ANCOVA Model

Predictor	F	P	Partial η2	B (95%CI)
Group (HC vs PD)	6.087	0.016	0.081	0.080 (0.015–0.145)
Years of education	3.379	0.070	0.047	0.010 (− 0.001–0.020)
Alcohol use (yes or no)	3.238	0.076	0.045	− 0.066 (− 0.139–0.007)
Age (years)	2.551	0.115	0.036	− 0.003 (− 0.008–0.001)

B: Unstandardized regression coefficient; CI: confidence interval; partial η²: partial eta-squared (effect size measure)

Correlation of DTI-ALPS index with clinical features

The heatmap analysis depicting correlations between demographic characteristics, neuroimaging metrics, and clinical symptoms (motor/non-motor) in Parkinson's disease patients is presented in Fig. 4. The computed imaging index exhibited significant negative correlations with both MDS-UPDRS Total scores (r = − 0.418, p = 0.005, BCa bootstrap 95% CI [− 0.628, − 0.171]) and MDS-UPDRS Part III scores (r = − 0.424, p = 0.005, BCa bootstrap 95% CI [− 0.644, − 0.169]). Separately, MDS-UPDRS Part III scores demonstrated no significant correlations with patient age (p = 0.084) or disease duration (p = 0.351). The computed imaging index showed no significant correlations with cognitive function (r = 0.265, p = 0.086), depressive symptoms (r = 0.013, p = 0.934), anxiety levels (r = − 0.273, p = 0.077). Separately, the computed imaging index suggested a trend toward a negative correlation with advancing age (r = − 0.334, p = 0.029, BCa bootstrap 95% CI [− 0.616, 0.008]).

Fig. 4.

The heatmap analysis depicting correlations between demographic characteristics, neuroimaging metrics, and clinical symptoms (motor/non-motor) in Parkinson’s disease patients

To determine the effect of the radiographic measure on motor symptoms in PD, multiple linear regression analysis was performed. The analysis demonstrated that after controlling for age, alcohol consumption history, and disease duration, the radiographic measure remained a significant negative predictor of motor symptom severity (β = − 0.420, p = 0.010; B = − 42.45, BCa bootstrap 95% CI [− 79.17, − 9.91]). The variable independently explained 14.6% of the variance in motor symptoms (ΔR² = 0.146, p = 0.008). All covariates demonstrated VIF < 1.31, indicating good model stability and absence of multicollinearity interference (see Supplementary Table 2).

Discussion

This study employed DTI-ALPS imaging to investigate differences in waste-clearance system activity between the PD group and the healthy control group. The diffusion tensor imaging parameters applied during scanning were configured based on established protocols from methodological references [25, 26], achieving an optimal balance between image data quality and acquisition efficiency. A b-value of 1000 s/mm² was selected for imaging, as this value adequately captures microstructural features of white matter tracts while preventing excessive signal attenuation associated with higher b-values.

The findings of this study indicate that, after adjusting for covariates including age, educational level, and alcohol consumption history, the DTI-ALPS values were significantly higher in healthy control subjects relative to individuals with Parkinson's disease. Further correlational assessment showed an inverse relationship between this functional parameter and the degree of motor impairment in the patient cohort. Multiple regression modeling demonstrated that the DTI-ALPS index independently accounted for 14.6% of the variance in motor symptoms (ΔR² = 0.146, p = 0.008). These results suggest that a reduced computed imaging index may be more strongly associated with motor dysfunction in PD, whereas no significant associations were found with the non-motor symptoms we assessed.

Consistent evidence from neuroimaging studies has indicated that patients with Parkinson’s disease commonly exhibit impaired glymphatic function, which is evidenced by a marked reduction in this functional parameter relative to healthy controls [27, 28]. Furthermore, glymphatic dysfunction in PD has been reported to be more severe than essential tremor [29], suggesting that this alteration may not be a mere consequence of movement disorders and could be more specifically linked to the pathophysiology of PD.

The current findings align with previously reported results. This functional parameter, as a non-invasive assessment metric, provides an indirect measure of the clearance pathway's efficacy in draining toxic byproducts of metabolism. In this study, the functional index in patients with Parkinson's disease was significantly lower, suggesting a decline in overall glymphatic function and potentially impaired diffusion along perivascular spaces.

Earlier investigations have observed an inverse relationship between this imaging metric and scores on the UPDRS-III motor subscale [30, 31]. The present findings further support this observation, confirming the negative association between this neuroimaging indicator and parkinsonian motor symptoms. In contrast to previous studies, the present investigation extends previous findings by providing a multivariate analysis of the association between this radiographic measure and parkinsonian motor features. Multiple linear regression analysis, controlling for age, drinking history, and disease duration, demonstrated that this MRI-derived metric remained a significant negative predictor of motor symptom severity. Our results indicate that as motor impairments progress in Parkinson’s disease, this radiographic measure decreases, suggesting impaired efficiency of the drainage network in eliminating toxic byproducts of metabolism. These findings add to the growing body of evidence that implicates aberrant perivascular fluid dynamics and glymphatic dysfunction in the pathological mechanisms of Parkinson's disease.

According to cerebrospinal fluid dynamics theory, the brain's waste-clearance system and the meningeal lymphatic drainage system operate in coordination to maintain fluid homeostasis within CNS. As the dural lymphatic vessels serve as the primary efflux pathway for cerebrospinal fluid exiting the brain parenchyma, impairment of this downstream pathway may cause flow obstruction, thereby adversely affecting the function of the upstream glymphatic system [32–34].

Animal experiments have shown that surgical ligation of meningeal lymphatic vessels in mouse models can lead to impaired clearance of α-synuclein from the CNS, accompanied by declines in autonomous motor and coordination functions. This suggests that impairment of the brain’s waste-clearance system may exacerbate parkinsonian motor deficits through disrupted AQP4-mediated clearance pathways [20, 35]. Furthermore, clinical investigations [35] have shown a marked decline in meningeal lymphatic drainage efficiency in individuals with idiopathic Parkinson’s disease compared to healthy subjects, indicating that impaired the CNS clearance and perivascular accumulation of α-syn may contribute to disease progression.

Animal studies have shown that the specific knockout of the aquaporin-4 gene results in enhanced pathological accumulation of α-syn in the perivascular spaces of the murine CNS, along with behavioral deficits such as impaired motor coordination [19]. This observation corroborates the mechanism whereby the glymphatic pathway relies on AQP4-mediated transport to clear neurogenic macromolecular metabolites, including α-syn. When this clearance system is impaired—for instance, by toxic accumulation due to aberrant α-syn aggregation—it may subsequently contribute to analogous parkinsonian motor features.

Current evidence indicates that the pathogenesis of PD may involve mechanisms of oxidative stress response and neuroinflammatory activation mediated by abnormal α-synuclein aggregation [36]. Animal studies have demonstrated that interventions with antioxidant and anti-inflammatory agents in mouse models of PD can attenuate the progression of motor deficits, indicating that neuron-derived inflammatory factors and oxidative stress may collectively participate in the onset and advancement of the disorder [37, 38].

The findings of this study complement previous animal experimental results. Although AQP4 expression was not directly measured, substantiation from established animal studies suggests that dysfunction of the AQP4-dependent perivascular clearance network could be associated with elevated pathological aggregation of α-syn, which could subsequently trigger neuroinflammation and neuronal damage, forming a positive feedback loop that exacerbates Parkinson’s disease pathology. The current results indicate that aberrant perivascular fluid dynamics may underlie certain pathological features of PD, and that impaired glymphatic clearance might be involved in the pathogenesis of the disorder.

The link between parkinsonian non-motor features and the clearance network, as assessed by this radiographic measure, remains inconclusive. The current study's results are in line with the observations made by Colin D. McKnight et al. [29], who did not find any marked link between this imaging parameter and cognitive decline as examined using MMSE. It should be noted that the MMSE, as a global cognitive screening instrument, has inherent sensitivity limitations—particularly in detecting early or mild deficits in cognitive domains characteristically affected in PD, such as executive dysfunction and visuospatial impairment [39]. This methodological constraint may partly explain the lack of significant associations in some studies. Notably, Ting Shen et al. [27] observed a clinically relevant reduction in this imaging biomarker in patients, which correlated negatively with cognitive scores—a finding that diverges our results. Such discrepancies underscore the need for further large-scale studies.

This investigation has multiple limitations. Firstly, given its cross-sectional nature, the study is unable to establish a causal link between glymphatic system dysfunction and α-synuclein deposition, nor can it entirely rule out the likelihood of residual confounding variables. Secondly, the restricted sample size and the less than comprehensive control of potential confounders may compromise the statistical robustness. Hence, these results necessitate validation through larger-scale studies incorporating longitudinal follow-up to track the progression of the disease.

Conclusion

This study suggests that DTI-ALPS imaging could be used to detect a decline in glymphatic system activity of patients, and that this functional indicator is significantly correlated with the intensity of motor symptoms. These insights further support the argument that glymphatic dysfunction may be a part of the pathophysiological process of the disorder, providing clinical neuroimaging evidence that could help elucidate disease mechanisms. The imaging technique may serve as an auxiliary tool for evaluating this disease and could potentially inform future research on therapeutic strategies targeting the glymphatic system. This imaging index could be explored as a potential neuroimaging biomarker for evaluating parkinsonian motor impairment and warrants further investigation in future studies.

Author contribution

M.X. wrote the main manuscript text, designed and conducted the experiment. Y.R. and W.K. provided administrative support. Y.R. and D.J. provided study materials or patients. M.X., J.Y. and H.L. collected and assembled data. M.X., Z.J., X.B. and G.W. undertook data analysis and interpretation. All authors read and approved the final submission.

Funding

This work was supported by the Henan Province Medical Science and Technology Research Program Joint Construction Project [grant numbers LHGJ20230505]; the Henan Province Higher Education Key Scientific Research Project [grant number 24B320017].

Data availability

The datasets used and/or analyzed during this study are available from the corresponding author on reasonable requests.

Declarations

Ethics approval and consent to participate

The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was approved by Ethics Committee of the First Affiliated Hospital of Xinxiang Medical University (Approval No. EC-024-106), and informed consent was provided by all individual participants.

Competing interests

The authors declare no competing interests.